On th e influence of dopamine-r elat ed genetic vari ati on on dopamine-r elat ed disorders
Oll e Bergman 2009
Department of Pharmacology Institute of Neuroscience and Physiology The Sahlgrenska Academy at University of Gothenburg
Sweden
Printed by Intellecta Infolog AB, Gothenburg
Olle Bergman 2009 ISBN 978-91-628-7938-9 http://hdl.handle.net/2077/21077
ABSTRACT
Rati onal e Dopamine synthesizing neurons are involved in a wide variety of functions.
The most prominent dopamine pathways originate in the midbrain. The development, function and survival of these dopaminergic neurons are under the influence of numerous transcription and neurotrophic factors. Subtle differences in the genes encoding these factors may be of importance for several psychiatric and neurodegenerative disorders.
LMX1A, LMX1B and PITX3 are transcription factors that are essential for the
development, specification and survival of midbrain dopaminergic neurons. BDNF is a neurotrophic factor involved in neurodevelopmental processes including differentiation and survival of dopaminergic neurons. Another protein of importance for dopaminergic neurotransmission is the dopamine transporter (DAT) that mediates reuptake and inactivation of extracellular dopamine and is hence of fundamental importance in regulating dopamine transmission. The specific aim of this thesis was to investigate the possible influence of polymorphisms in these dopamine-related genes on dopamine- related disorders, i.e. Parkinson’s disease (PD), attention-deficit/hyperactivity disorder (ADHD), social anxiety disorder (SAD) and schizophrenia. OO b servati ons Three single nucleotide polymorphisms (SNPs) in LMX1A and one in LMX1B were associated with PD. After splitting for gender, six SNPs were associated with PD in women and four in men (Paper I). Two SNPs in PITX3 were associated with PD in patients with an early age of onset when compared either to controls or to PD patients with late onset (Paper II). One of the PITX3 polymorphisms was also associated with schizophrenia, as were two polymorphisms in LMX1A, and one SNP in LMX1B (Paper III). We assessed longitudinal, quantitative phenotypes of hyperactivity-impulsivity and inattention, and found that the Met allele of the Val66Met polymorphism in the BDNF gene was associated with increased persistent hyperactivity-impulsivity symptoms as well as with increased age-specific inattention symptoms (Paper IV). The amygdala, essential for detection of biologically relevant stimuli and fear generation, is under excitatory influence of dopamine. Positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) were used to investigate if a variable number of tandem repeat (VNTR) polymorphism in the DAT gene (SLC6A3) influences amygdala function during processing of aversive emotional stimuli in SAD patients and healthy controls,
respectively. The 9-repeat allele was associated with significantly increased amygdala activity, as assessed with PET, across tests (i.e. public speaking, processing of angry and neutral faces) in SAD patients, but with decreased amygdala activity in controls.
Moreover, 9-repeat carriers, regardless of diagnosis, displayed augmented amygdala reactivity, i.e. a greater activation, of the left amygdala in response to angry compared to neutral faces. Blood oxygen level-dependent (BOLD) fMRI was used to assess healthy volunteers, and in line with the results from the PET study, 9-repeat carriers displayed higher reactivity of the left amygdala in response to angry faces, compared to neutral geometric shapes (Paper V). CConclusi ons All of the studies were based on a priori hypotheses regarding the possible relationship between the genes and the disorders under investigation. Some of the associations reported in this thesis have not been described earlier, others have been confirmed in independent samples, whereas in some cases, earlier studies have been inconclusive. In summary, our results support the notion that variation in dopamine-related genes is of importance for dopamine-related disorders and amygdala function.
This the sis is b ased on the following papers, which will be referred to in the t ext by their Roman numerals:
I. PITX3 polymorphism is associated with early onset Parkinson’s disease. OOll e Bergman, Anna Håkansson, Lars Westberg, Kajsa Nordenström, Andrea Carmine Belin, Olof Sydow, Lars Olson, Björn Holmberg, Elias Eriksson and Hans Nissbrandt. Neurobiology of Aging (2008) Apr 16 (Epub. ahead of print).
II. Do polymorphisms in transcription factors LMX1A and LMX1B influence the risk for Parkinson’s disease? OOll e Bergman, Anna Håkansson, Lars Westberg, Andrea Carmine Belin, Olof Sydow, Lars Olson, Björn Holmberg, Laura Fratiglioni, Lars Bäckman, Elias Eriksson, Hans Nissbrandt. Journal of Neural Transmission (2009) 116:333–338.
III. Polymorphisms in dopamine-related transcription factors LMX1A, LMX1B and PITX3 are associated with schizophrenia. OOll e Ber gman, Lars Westberg, Lars- Göran Nilsson, Rolf Adolfsson and Elias Eriksson. Preliminary manuscript.
IV. Association of brain-derived neurotrophic factor polymorphism with the developmental course of attention-deficit/hyperactivity disorder. OOll e Bergman, Lars Westberg, Paul Lichtenstein, Elias Eriksson and Henrik Larsson.Submitted manuscript.
V. Amygdala function is associated with a dopamine transporter gene polymorphism in patients with social anxiety disorder and healthy controls. OO ll e Bergman, Fredrik Åhs, Tomas Furmark, Lieuwe Appel, Clas Linnman, Vanda Faria, Stephen B. Manuck, Robert E. Ferrell, Ahmad Hariri, Susanne Henningsson, Mats Fredrikson, Elias Eriksson,and Lars Westberg. Submitted manuscript.
Reprints were made with kind permission from the publisher
TABLE OF CONTENTS
LIST OF ABBREVIATIONS 8 INTRODUCTORY COMMENT 9 GENETICS AND HERITABILITY 10
DNA, RNA AND PROTEINS 10
CLASSES OF GENETIC VARIATION 10
CROSSOVER AND RECOMBINATION 12
LINKAGE AND LINKAGE DISEQUILIBRIUM 13
CONCEPTS OF GENETIC STUDIES 14
TRAITS, PENETRANCE AND HERITABILITY 14
LINKAGE STUDIES 15
CANDIDATE GENES 16
ASSOCIATION STUDIES 16
DOPAMINE 18
DOPAMINE SYNTHESIS, TRANSMISSION AND RECEPTORS 18
DOPAMINE CELL GROUPS IN THE CNS 19
TRANSCRIPTIONAL CONTROL OF DOPAMINERGIC NEURON DEVELOPMENT 20
EARLY EMBRYOGENESIS 20
FORMATION OF THE ISTHMUS AND MDDA NEURONAL FIELD 21
DIFFERENTIATION OF MDDA NEURONS REQUIRES LMX1A 22
DEVELOPMENT OF POSTMITOTIC MDDA NEURONS INVOLVES LMX1B AND PITX3 23 PARKINSON’S DISEASE 25
BACKGROUND, SYMPTOMATOLOGY AND TREATMENT 25
PATHOPHYSIOLOGY AND ETIOLOGY OF PARKINSON’S DISEASE 26
HEREDITY AND GENETIC FACTORS IN PARKINSON’S DISEASE 27
SCHIZOPHRENIA 28
BACKGROUND, SYMPTOMATOLOGY AND TREATMENT 28
PATHOPHYSIOLOGY AND ETIOLOGY OF SCHIZOPHRENIA 29
HEREDITY AND GENETIC FACTORS IN SCHIZOPHRENIA 31
ATTENTION DEFICIT/HYPERACTIVITY DISORDER 33
BACKGROUND, SYMPTOMATOLOGY AND TREATMENT 33
PATHOPHYSIOLOGY AND ETIOLOGY OF ADHD 34
HEREDITY AND GENETIC FACTORS IN ADHD 36
SOCIAL ANXIETY DISORDER 37
BACKGROUND, SYMPTOMATOLOGY AND TREATMENT 37
PATHOPHYSIOLOGY AND ETIOLOGY OF SOCIAL ANXIETY DISORDER 38
HEREDITY AND GENETIC FACTORS IN SOCIAL ANXIETY DISORDER 39
THE AMYGDALA 39
ANATOMY AND CONNECTIONS OF THE AMYGDALA 39
AMYGDALA FUNCTIONS 40
INFLUENCE OF DOPAMINE ON AMYGDALA FUNCTION 41
AIMS OF THIS THESIS 43 RESULTS AND DISCUSSION 45
PAPER I. PITX3 POLYMORPHISM IS ASSOCIATED WITH EARLY ONSET PARKINSON’S DISEASE 45 PAPER II. DO POLYMORPHISMS IN TRANSCRIPTION FACTORS LMX1A AND LMX1B
INFLUENCE THE RISK FOR PARKINSON’S DISEASE? 47
PAPER III. POLYMORPHISMS IN DOPAMINE-RELATED TRANSCRIPTION FACTORS LMX1A, LMX1B AND PITX3 ARE ASSOCIATED WITH SCHIZOPHRENIA 48 PAPER IV. ASSOCIATION OF BRAIN-DERIVED NEUROTROPHIC FACTOR POLYMORPHISM WITH THE DEVELOPMENTAL COURSE OF ATTENTION-DEFICIT/HYPERACTIVITY DISORDER 50 PAPER V. AMYGDALA FUNCTION IS ASSOCIATED WITH A DOPAMINE TRANSPORTER GENE POLYMORPHISM IN PATIENTS WITH SOCIAL ANXIETY DISORDER AND HEALTHY CONTROLS 51
CONCLUDING REMARKS 54
APPENDIX: SUBJECTS AND METHODS 56
SUBJECTS 56
METHODS FOR GENOTYPING AND SEQUENCING 58
NEUROIMAGING METHODS 60
STATISTICS 62
ACKNOWLEDGEMENTS 65 REFERENCES 66
LIS T OF ABBREVIATIONS
CNS Central nervous system
mdDA Mesodiencephalic dopaminergic
VTA Ventral tegmental area
SNc Substantia nigra pars compacta
RRF Retrorubral field
FGF2/8 Fibroblast growth factor 2/8
Shh Sonic hedgehog homolog
Otx2 Orthodenticle homeobox 2
Gbx2 Gastrulation brain homeobox 2
Lmx1a/b LIM homeobox transcription factor 1 alpha/beta
Pitx3 Paired-like homeodomain 3
En1/2 Engrailed homeobox 1/2
Pax2/5 Paired box 2/5
Foxa1/2 Forkhead box A1/2
Tgf/ Transforming growth factor alpha/beta Nurr1 (NR4A2) Nuclear receptor subfamily 4, group A, member 2
PD Parkinson’s disease
SAD Social anxiety disorder
ADHD Attention-deficit/hyperactivity disorder
Msx1 Msh homeobox 1
Nkx6.1 NK6 homeobox 1
PET Positron emission tomography
fMRI Functional magnetic resonance imaging
rCBF Regional cerebral blood flow
GABA Gamma-Aminobutyric acid
AMPA -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid
NMDA N-methyl-D-aspartic acid
NO Nitric oxide
COMT Catechol-O-methyltransferase 5-HTTLPR (SLC6A4) Serotonin transporter gene
SLC6A3 (DAT1) Dopamine transporter gene ADRB1 1-adrenergic receptor gene
DAO D-amino-acid oxidase
DRD1/2/4 Dopamine receptor D1/2/4
NRG1 Neuregulin 1
AKT1 v-akt murine thymoma viral oncogene homolog 1 BDNF Brain-derived neurotrophic factor
DISC1 Disrupted in schizophrenia 1
DTNBP1 Dystrobrevin binding protein 1
DNA Deoxyribonucleic acid
RNA Ribonucleic acid
LD Linkage disequilibrium
GWAS Genome-wide association study
VNTR Variable number of tandem repeat
SNP Single nucleotide polymorphism
INTROD UCTO RY COMM ENT
"How extremely stupid not to have thought of that!" was Thomas Henry Huxley’s response to Charles Darwin’s theory of natural selection (1838). In hindsight the idea of natural selection and its requirements seem obvious; more individuals are born than can survive and reproduce, these individuals vary in their ability to survive and reproduce and this variety is partly heritable. Importantly, this variety also allows for traits that under certain conditions can become deleterious and cause disease.
Whereas the disorders investigated in this thesis are very different from one another, all have a large genetic component and are assumed to be associated with the
neurotransmitter dopamine. Dopamine is involved in a variety of brain functions,
including cognition, locomotor activity and emotion (Vallone et al, 2000). Because of its importance to the pathology of several brain disorders such as schizophrenia, attention- deficit/hyperactivity disorder (ADHD) and Parkinson’s disease (PD), dopamine has been a major target of research ever since its function as a neurotransmitter was discovered by Arvid Carlsson and co-workers in the late 1950s (Carlsson and Waldeck, 1958; Iversen and Iversen, 2007).
Since the first human genome sequence was published in 2001 (Lander et al, 2001;
Venter et al, 2001), efforts to link human genetic variation with commonly occurring disorders have been intensified. We all carry two versions of each chromosome and consequently have two versions of most genes (with the exception of genes on sex chromosomes). These two versions (alleles) are almost identical, but the devil is in the details; subtle genetic differences are the foundation of all psychiatric and
neurodegenerative disorders.
Findings from twin studies suggest that genetic factors play an important role in the pathogenesis of three of the disorders investigated in this thesis (schizophrenia, social anxiety disorder (SAD) and ADHD), and some importance also for the fourth (PD).
Unfortunately, nature is not bound to our diagnostic nomenclature, and the genetic heterogeneity of these disorders is substantial. Moreover, very few of these genetic factors result in changes in protein structure and function. More often, aspects that have
relatively subtle biological effects, like gene regulation, may be implicated. Importantly, whereas patients do not inherit an illness per se, they may inherit variants leading to an altered brain development, which, in combination with environmental factors, may increase the risk of developing a disorder later in life. Consequently, these disorders are genetically complex and identifying robust susceptibility genes have so far proven to be very difficult. Nevertheless, identifying the underlying genetic components of these disorders will be necessary to fully understand their pathophysiology, and will hopefully lead to the development of more rational treatments than those available today.
The aim of this thesis has been to investigate how and if dopamine-related genetic variants influence the susceptibility to schizophrenia, PD, symptoms of ADHD and amygdala function in SAD patients and healthy controls, respectively.
GENETICS AND HERI TABILITY DNA, RNA and prot eins
The human genome is made up of deoxyribonucleic acid (DNA), consisting of a sugar called deoxyribose, phospate groups and four different nucleotide bases: adenine (A), cytosine (C), guanine (G) and thymine (T). A and G are purines while C and T are pyrimidines. The DNA components are connected to each other with strong covalent bonds, forming a long single strand with two ends, called 5 and 3 (Lehman, 1974).
Weaker hydrogen bonds pair A with T and C with G, connecting two single strands to form a double helix of DNA (Watson and Crick, 1953). There are normally 46 DNA molecules, called chromosomes, consisting of 22 homologous pairs of autosomes and one pair of sex chromosomes, in the cell nucleus under normal cell conditions. One
chromosome in each pair is from the mother and the other from the father. Double stranded DNA is cleaved prior to replication and a complementary strand is created for each strand of DNA. Exons are parts of the DNA strand that encode genes. These are normally separated by introns, which are non-coding sequences. Regulatory sequences called promoter regions are normally located upstream of any given gene. Promoter regions contain special nucleotide sequences (i.e. motifs) that are recognized by a group of proteins called transcription factors. These proteins bind to promoter regions and participate in the control of gene expression. Whereas it has previously been assumed that promoter regions are usually located upstream from the 5' end of a gene, mapping studies found that in fact only 22% of transcription factor binding sites were located upstream of the 5' ends, whereas 36% actually lied within gene boundaries, often in noncoding areas (Cawley et al, 2004; Martone et al, 2003). The 3 untranslated region (3
UTR) is located downstream of a gene and contains regions which influence RNA stability and translation (Mill et al, 2002; Miller and Madras, 2002), as well as sites that are partially complementary to noncoding microRNAs. Different cell types express distinct combinations of microRNAs, which may regulate cell-specific target genes (Krek et al, 2005; Lai, 2002). The haploid (i.e. one copy of each chromosome) human genome consists of about 3.3 billion base pairs and harbours an estimated 20.000 to 30.000 protein-coding genes (2004; Stein, 2004). In fact only 1.1% of the genome is believed to consist of exons, whereas 24% is intronic, the remaining 75% is intergenic DNA (Lander et al, 2001; Venter et al, 2001). The parts of the genome encoding genes are transcribed to complementary RNA that subsequently becomes mRNA after non-translated
sequences have been removed in a process called post-transcriptional processing. Mature mRNA contains so-called codons (i.e. nucleotide trios) that are translated into amino acids, which combine to form proteins (Burton et al, 2005).
Classe s of genetic vari ati on
Small individual differences in the genome are the foundation of all common hereditary disorders, including the psychiatric and neurodegenerative ones. Several different forms of genetic variation have been identified; the most common are the single nucleotide polymorphisms (SNPs), which represent variation in a single nucleotide. A SNP can be defined as a locus (unique chromosomal location) in the genome at which two (or sometimes more) nucleotide bases can occur, all at a frequency of 1% or more.
Since we have two versions of each autosomal chromosome and consequently two versions of every gene, a polymorphic locus with two potential alleles (A and a) has three
possible genotypes: two homozygous (AA and aa) and one heterozygous (Aa). The Hardy-Weinberg principle states that both allele and genotype frequencies in a
population are in equilibrium, which can be calculated with the equation p2+2pq+q2=1, where p and q indicate frequency of allele A and a, respectively. However, this may not be true if the studied population is influenced by factors like selection, limited
population size, random genetic drift, non-random mating, mutations etc. Since one or more of these factors are always present in real life, Hardy-Weinberg equilibrium simply provides a baseline against which genetic variation can be measured.
It has been estimated that the human genome contains more than ten million SNPs, about seven million of these occurring with a minor allele frequency (MAF) of over 5%
(2003; Kruglyak and Nickerson, 2001). In addition to these common SNPs there are an inestimable number of rare single nucleotide mutations, in some cases occurring in only one family or individual. However, the greater part of the genetic variation that occurs between any two individuals is located at positions with variants that are common in the population as a whole (Frazer et al, 2009).
SNPs have until recently been thought to affect disease susceptibility mainly by altering the DNA sequence in an exon so that the amino-acid coding is changed. So-called non- synonymous SNPs give rise to novel codons that specifies an alternative amino-acid or changes the code for an amino-acid to that for a stop signal. Most susceptibility SNPs are however noncoding, and while some may function as markers for nonsynonymous variants, others are likely to alter the expression of a gene by themselves. Numerous studies have shown that promoter polymorphisms can influence transcriptional activity of a gene (Caspi et al, 2003; Greenwood and Kelsoe, 2003; Laws et al, 2002; Wilson et al, 1997). Similarly, intronic SNPs can also affect transcription, or alter splicing or mRNA stability, and hence change the relative quantity and proportions of isoforms
(Greenwood et al, 2003; Tokuhiro et al, 2003; von Ahsen and Oellerich, 2004). In addition, SNPs in the 3' UTR have also been shown to increase the susceptibility of certain disorders (Ueda et al, 2003) by altering mRNA stability or by being located in motifs for microRNAs that inhibit translation of a gene (Mill et al, 2002; Miller et al, 2002). Synonymous (conservative) exonic SNPs alter DNA sequence but do not change the amino-acid sequence. Synonymous SNPs are usually considered functionless, but some are under natural selection, indicating that they may cause disease, most likely by altering mRNA structure and translation (Chamary et al, 2006; Duan et al, 2003;
Kimchi-Sarfaty et al, 2007; Komar, 2007; Shen et al, 1999).
Other forms of polymorphisms include repeats. Microsatellite repeat polymorphisms include mono- di-, tri-, tetra-, and pentanucleotide tandem repeats that are dispersed in most chromosomes. There are on average one dinucleotide repeat occurring on every 30.000 bases in the human genome (Stallings et al, 1991). Minisatellite repeats are also known as variable number tandem repeats (VNTRs). They contain a repeating unit that is usually around 30 to 50 bp in length with a conserved core sequence of 10 to 15 bp, and occur at around 1000 sites in the genome (Doggett, 2001).
The parts of the human genome sequence that do not encode genes sometimes contain transposons (also called insertion sequences), believed to be the remains of intracellular parasites from our evolutionary history. These elements are excised from one site in the human genome sequence and integrated into another site, usually located less than 100
kb from the original site (Muotri et al, 2007). Retrotransposons constitute another form of genetic variation in humans, which are similar to retroviruses, and are elements that have been transcribed into RNA, reverse transcribed and subsequently reintegrated into the genome, thus duplicating the element (Kazazian, 2004). Genetic structural variation, including large (more than 1 kilobase pair (kbp)) insertions, deletions, inversion of sequences, block substitutions and duplications (also called copy-number variations) are common in the human genome (Feuk et al, 2006; Kidd et al, 2008). When the human genome sequence was published in 2001, it was thought that genetic differences accounted for approximately 0.1% of the sequence and consisted mostly of SNPs (Lander et al, 2001; Venter et al, 2001). Structural variation is today believed to
encompass around 1% of any given persons genome and have received increasing interest in later years after it has become increasingly clear that complex disorders cannot be easily explained by SNP variation alone (Frazer et al, 2009; Kidd et al, 2008). As a result, appropriate methods of detecting structural variants and their association with complex traits have started to appear (Conrad et al, 2006; Conrad et al, 2009; Hastings et al, 2009; McCarroll and Altshuler, 2007; McCarroll et al, 2008; Redon et al, 2006). These studies indicate that structural variants accounts for approximately 20% of genetic variation in humans, hence underlining their importance in future studies of human disease.
Importantly, altered expression of a gene does not necessarily occur because of genetic variants in that specific gene. Susceptibility genes may encode proteins that influence transcriptional pathways, which in turn may lead to compensatory or secondary changes.
Finally, the relationship between genotype and disorder can be complicated by epigenetic factors, which are heritable factors that do not cause sequence variation, (e.g. alterations in DNA methylation and chromatin structure) (Jaenisch and Bird, 2003; Robertson and Wolffe, 2000).
Paper I, II and III address syn on ymous SNPs in transcr iption fac tor genes, Paper IV addresses a non- syn onym ous SNP in BDNF, and Paper V addresses a repeat polymorphism in the SLC6A3 gene.
Crossover and rec ombinati on
During meiosis cells divide into four gametes, each containing a haploid genome. These gametes may fuse with another gamete in sexually reproducing animals. In the course of this gamete formation, the gamete receives a combination of the two homologous chromosomes. Crossover connections during meiotic division create a possibility for the paternal chromosomes to exchange genetic material at crossover sites called chiasmata.
DNA segments are exchanged at these sites at corresponding positions along pairs of homologous chromosomes by symmetrical breakage and crosswise rejoining. Crossovers are more likely to occur in some parts of the genome, called hotspots, where it is
believed that conservation is less important (Kauppi et al, 2004). The longer the distance between two genes, the higher the probability of an odd number of crossovers, which causes so-called recombination. Recombination has the ability to split alleles that are located together on a common parental chromosome and to position alleles that originally came from different grandparents on the same chromosome. This creation of new haplotypes increases genetic variability and consequently exerts an important evolutionary influence (Dawn Teare and Barrett, 2005).
Another form of recombination is gene conversion, which involves a one-way transfer of DNA from a "donor" sequence to a highly homologous "acceptor" sequence (Chen et al, 2007). Gene conversion occurs in both meiotic and somatic cell division in humans, and can be defined as the transfer of information between alleles or loci without crossover, e.g. the meiotic products of an individual carrying Aa at a locus may be AAAa or aaaA instead of AAaa, i.e., the A allele has been converted into the a allele or vice versa (Hellenthal and Stephens, 2006).
Linkage and linkage disequilibrium
Linkage and linkage disequilibrium (LD) are two important elements in genetic
epidemiology. Both concepts measure a correlation, or co-segregation, between genetic markers. Whereas linkage measures co-segregation of loci in a pedigree, LD measures co- segregation of alleles in a population. When a new SNP mutation occurs close to an older one, both SNP alleles are often transmitted together. One locus is in linkage with another locus on the same chromosome if they are transmitted together from parent to offspring more often than should be expected in the case of independent inheritance; that is, if recombination occurs between them with a probability of less than 50%. Alleles at two or more loci are in LD if they are transmitted together in the same haplotype more often than expected across the total population. Each time that recombination takes place between the loci in the total population, LD between these loci becomes weaker, and is only preserved if the two loci are located very close to each other. If two loci are in strong LD, little recombination has taken place between them in the past. Two loci in LD will always be linked, but the reverse is hence not necessarily true (Dawn Teare et al, 2005).
LD is not a quantitative measurement and there is no natural scale for measuring it.
There are several LD measures describing the statistical association between alleles at different loci, the most common being D' and r2, both of which are two-locus measures but may be used for different purposes (Mueller, 2004). Central for both these
measurements is the linkage disequilibrium coefficient D. If p equals the allele frequency at two loci (1 and 2), each containing two alleles, A and a at locus 1 and B and b at locus 2, the frequency of A and B occurring together on the same chromosome is pAB and the covariance between the two loci is D = pAB - pApB, where pApB is the expected value of pAB
when allelic association is missing. The linkage disequilibrium coefficient D is constrained in the value it may take so in order to compare LD between loci with differing allele frequencies, D' and r2 is used (Mueller, 2004). D' is determined as the ratio between D and Dmax, always ranging between 0 and 1 (1 indicating high LD). The squared coefficient of determination r2 is determined as D2 divided by the product of all allele frequencies: r2 = D2/{pApB (1-pA)(1-pB)}. r2 is similar to D' in that it ranges from 0 to 1 (where 1 indicates high LD). However r2 scales D by the standard deviations of the allele frequencies at the two loci, in contrast to D' that scales D by its maximum value.
One weakness of using D' is that it tends to be overblown; D' may equal 1 while r2 at the same time is much lower. Only if r2 equals 1 do two alleles always occur together on a haplotype, resulting in a maximum of two possible haplotypes. Consequently the identity of an allele at locus 1 can be determined by information regarding the allele at locus 2. D' is more useful to assess probability for evolutionary recombination in a population,
whereas r2 is more helpful in association studies (Balding, 2006; Devlin and Risch, 1995;
Mueller, 2004).
The alleles of SNPs located close to each other are often correlated with one another. A series of alleles at linked loci on a single chromosome is called a haplotype (Daly et al, 2001). The LD between SNPs varies from place to place in the genome and between different populations. Instead of genotyping all SNPs in a non-recombining haplotype block it is possible to use a number of haplotype tag SNPs (tSNPs) as markers for disease mapping (Johnson et al, 2001). These tSNPs capture the genetic information in each haplotype block, making it possible to genotype fewer SNPs, which saves both money and time. However, LD between SNPs varies from place to place in the genome and between different populations.
The HapMap project was launched in 2001 with the aim to map haplotype block structure in different ethnic populations and subsequently define tSNPs for each block (Barrett et al, 2005). Data from the HapMap project suggest that a vast majority of SNPs with a minor allele frequency (MAF) of at least 5% could be reduced to only 550.000 LD blocks for individuals of European descent (r2 > 0.8). Subsequently, information on more than 80% of SNPs across the genome can be obtained by
genotyping tSNPs from each LD block (2003). It should be noted that the definition of haplotype blocks depend on tSNP density and that, for certain regions or populations, high frequency tSNPs or apparent LD block boundaries may not exist. Furthermore, susceptibility SNPs that are located in a recombinational hot spot are impossible to detect using haplotype-tSNPs (Kauppi et al, 2004).
CONCEPTS OF GENETIC STUDIES Trai t s, penetr anc e and heritability
Mendelian traits are named after Gregor Mendel (1822-1884) who studied inheritance of certain traits in plants and found that these follow particular laws. Mendel crossed white and purple pea flowers and discovered that the offspring was not a hybrid but rather a 3:1 ratio of purple and white flowers, respectively. He hypothesized that genes can be paired in three ways for each trait, AA, Aa and aa, where A represents a dominant trait, in this case the colour purple. The genotype of a purple flower can thus be either AA or Aa, while white flowers (white being the recessive trait) can only be carriers of the aa genotype. If an AA genotype carrier is crossed with a carrier of the aa genotype, all offspring will carry the Aa genotype and be purple because that is the dominant trait. If Aa genotype carriers are crossed with white flowers, on the other hand, half of the offspring will be white aa carriers and the other half will carry the Aa genotype and be purple.
Mendel summarized his findings in the law of segregation and the law of independent assortment. The law of segregation states that the paternal and maternal chromosomes get separated during meiosis and alleles are segregated into two gametes. The law of independent assortment states that alleles assort independently of one another during gamete formation. A Mendelian trait is thus one that is controlled by a single locus and shows a simple Mendelian inheritance pattern. In such cases, a mutation in a single gene can cause a disease that is inherited according to Mendel's laws. Under Mendelian
inheritance, only the fitness of the individual is important; the sole determinant of whether an allele will spread when it enters a population is whether the fitness of heterozygotes is greater than that of wild-type homozygotes (Hurst, 2009).
Complex traits do not follow Mendelian mode of inheritance. In complex traits under non-Mendelian inheritance, the faith of a new allele is determined both by the rate of transmission and the fitness of the organism. Complex disorders may depend on several susceptibility gene variants, and on combinations of genes interacting with environmental factors (Hurst, 2009). Finding solid genetic risk factors for psychiatric and other complex disorders have proven difficult. There are a number of possible explanations to this.
Complex traits are the result of variation in several different genes (i.e. locus
heterogeneity), with individual risk alleles usually explaining only 1-5% of the variation in a studied trait (Burmeister et al, 2008). Complex disorders are often based on a diagnosis rather than actual pathophysiological changes. It is hence possible that certain complex disorders comprise several separate etiologies caused by different susceptibility genes, which vary from patient to patient. Interaction between genes and between genes and environmental factors add to a situation where a specific allele increases the risk of developing a disease in one person but not in another (Cordell, 2009). In addition, some risk alleles display incomplete penetrance, i.e. a particular phenotype is not always expressed in a person with a particular genotype.
Linkage studies
Genetic linkage analysis can be used to identify genomic regions that include genes predisposing to disease. In parametric (model-based) linkage analyses, the co-segregation of linkage loci in pedigrees is analysed (Dawn Teare et al, 2005). Loci situated close to each other on the same chromosome segregate together more often than do loci on different chromosomes, which segregate together by chance only. As discussed above, recombination at meiosis is more likely the longer the distance between two loci on a chromosome. The recombination fraction measures the recombination between loci on a chromosome in offspring. When two loci are unlinked the recombination fraction is said to be 0.5. Linkage analysis measures deviation from this number and can be attained by genotyping linkage markers and studying their segregation in pedigrees. If one or more markers show signs of co-segregation, these are said to show linkage to the disease and are hopefully associated with a gene that may be responsible for the disease. Linkage markers are usually microsatellites (short regions of tandem repeats) or SNPs (1992). In linkage studies, DNA is collected from large extended families (or pedigrees) and several hundred or thousand (depending on resolution) evenly distributed DNA markers are analysed to see whether these markers co-segregate with disease in the pedigree.
Parametric linkage analyses have been successful in finding disease loci for disorders with a simple Mendelian inheritance. However, psychiatric and neurodegenerative disorders tend to be very complex in nature and to involve several genes with no clear mode of inheritance. Non-parametric (model-free) linkage analysis does not require specification of a disease model. The underlying principle is that, between affected relatives excess sharing of haplotypes that are identical by descent (IBD) in the region of a disease- causing gene is expected, and does not depend on the mode of inheritance. However, linkage analyses of complex disorders are usually only able to identify large regions that
often contain hundreds of genes, and are thus of limited usefulness in studies of complex traits (Dawn Teare et al, 2005; Prathikanti and Weinberger, 2005).
Candidate gene s
A candidate gene is a gene that is believed to play a causal role in a particular disease (Botstein and Risch, 2003; Cordell and Clayton, 2005). If the pathophysiology of the disease is known, defining candidate genes and determining which gene variants that predict who becomes ill may be quite straightforward. For most psychiatric and neurodegenerative disorders our definitive knowledge regarding pathophysiological mechanisms is sparse. However, for many conditions there are usually reasonable theories implicating e.g. certain neurotransmitters in the underlying biological
aberrations, prompting researchers to study the possible importance of genes regulating the transmitter in question as putative susceptibility genes for the investigated disorder.
For example, studies of disorders that have a pathophysiology tentatively involving monoamine transmitters have to a great extent focused on genes linked to monoamine neurotransmission.
Candidate genes may also arise from positional linkage studies. As discussed above, such studies may identify a certain area of a chromosome that is linked with a disorder; genes located in this specific area are subsequently candidates and may be the subject of association studies in order to determine whether they play a role in the etiology of the disorder being studied.
Associ ation studies
In association studies the relationship (or association) between a specific allele and a phenotype (e.g. a quantitative trait or a disease) is studied. Association studies may thus, for example measure whether an allele is more frequent in a patient population relative to a control population.
After having decided which genes to study, one needs to decide which candidate
polymorphisms in these genes to investigate. This decision is not an obvious one, since it is likely that many causal SNPs involved in common complex disorders will be non- coding, causing variation in gene regulation, expression or splicing. As discussed above, SNPs are inherited from parent to offspring in chromosome blocks containing many SNPs. Due to recombination at meiosis, the size of the DNA block containing the founder SNP shrinks over the years. Since all polymorphisms within a block are in linkage disequilibrium, and thus inherited together, all of them can function as markers for a disease causing SNP within the block. Thus, it is possible that an associated SNP has no causal role but is associated with a nearby causal polymorphism, copy-number variation or deletion (Hinds et al, 2006; Locke et al, 2006). As discussed above, the use of haplotype tSNPs, which capture the genetic variation of the full haplotype, may increase the power of an association study. Another option is to select polymorphisms of known functional importance, e.g. nonsynonymous SNPs that are likely to affect the function of the protein (Hirschhorn and Daly, 2005).
In addition to the investigation of candidate polymorphisms or tSNPs, association studies have also been used in fine mapping genetic loci initially detected by linkage
analysis. However, with the advance of genotyping methods it is now possible to scan a large numbers of SNPs at rapidly falling costs. This can be used to investigate
associations in thousands of loci at the same time. One may, for example, study all SNPs in a gene of interest, or even 500.000 or 1.000.000 SNPs throughout the entire genome, making a priori selection of candidate genes redundant.
However, there are drawbacks to this approach. Statistical analysis of several thousands of genotypes thus requires harsh correction for multiple testing, requiring effect sizes to be very large in order for p-values to be considered significant (Ziegler et al, 2008). This obviously is a problem when studying complex disorders that may be caused by many vulnerability polymorphisms, each exerting a modest effect, which interact to cause a certain disorder.
Replication in an independent population is usually the best way of confirming that an observed association is “true” rather than accidental. However, there are many reasons why a “true” association might not be replicated, other than that should be “untrue”.
First, allelic heterogeneity between different ethnic groups might result in different polymorphisms within the same gene to contribute to disease risk. Therefore,
comparisons should preferably be made within ethnically homogeneous subpopulations (Cordell et al, 2005).
Second, it is possible that the association is modified by other factors - genetic or environmental - that differ between studied groups (Wang et al, 2005).
Third, the disease-causing allele may be in LD with separate markers in different groups (Hinds et al, 2006; Locke et al, 2006).
Fourth, selection of cases may differ between studies due to the use of diverse diagnostic methods; as mentioned above, complex disorders often display clinical heterogeneity. It is therefore common to try to reduce phenotypic heterogeneity by studying traits that are less heterogeneous than most psychiatric diagnoses, such as a specific symptom of a disorder (e.g. inattention and hyperactivity-impulsivity, respectively, in ADHD
diagnosis), subpopulations (e.g. gender or age groups) or intermediate phenotypes (also called endophenotypes). An intermediate phenotype is a heritable phenotype that is associated with a disorder but can be measured independently of disease status (e.g.
amygdala activity during emotional stimuli in SAD patients). Intermediate phenotypes are thought to be closer to the pathogenic genotype than the clinical phenotype itself (Meyer-Lindenberg and Weinberger, 2006; Prathikanti et al, 2005).
In addition, other issues that might complicate replications of genetic associations include inadequate sample sizes, differences in statistical methods and the so-called winner's curse, which may lead to an overestimation of effect sizes, thereby handicapping replication (Zollner and Pritchard, 2007).
All papers included in this th esis ar e associati on studies. Paper I, II and III look at dichotomous tr aits (c ase/contr ol), whereas Paper IV and V in vestigate continu ous outc ome measur es. Paper IV illustrates the advantage of studying diagnostic subgroups, and Paper V the endop henotype approach.
DO PAMINE
Dopamine synthesis, transmission and rec e ptor s
Neurons in the CNS communicate mainly by neurotransmitters, which in turn can be divided into three main categories: amino acids, peptides and amine transmitters. The latter group includes the catecholamine dopamine (Carlsson et al, 1958; Hokfelt et al, 1987; Moore, 1993).
Like other catecholamines, the dopamine molecule structure has a core consisting of a benzene ring with two adjacent hydroxyl groups (catechol) and a single amine group.
The precursor for the synthesis of dopamine is the aromatic amino acid tyrosine, which is transformed into dopamine in two steps. The first reaction is catalysed by the rate-
limiting enzyme tyrosine hydroxylase (TH), which transforms tyrosine into L-3,4- dihydroxyphenylalanine (L-DOPA). A second step catalysed by the enzyme aromatic L- amino acid decarboxylase (AADC) leads to decarboxylation of L-DOPA to dopamine (Vallone et al, 2000).
Pacemaker-like membrane currents in dopamine neurons drive the spontaneous baseline activity (tonic firing) seen in dopaminergic neurons, which is the subject of strong GABAergic inhibition (Dewey et al, 1992). Phasic firing, on the other hand, is dependent on glutamatergic excitatory input, and displays a burst spike firing pattern, which triggers high amplitude synaptic dopamine release.
Following release dopamine is inactivated by reuptake transportation into pre-synaptic terminals via the dopamine transporter (DAT) (Goto et al, 2007; Grace and Bunney, 1984a, b). DAT levels are high in subcortical structures such as the striatum and amygdala, where it is located within the synapse, whereas levels are low in the cortex, where DAT localization is more distant from the site of release (Slifstein et al, 2008), as seen in studies on both rodents and monkeys (Lewis et al, 2001; Sesack et al, 1998).
The action of dopamine released into the synaptic cleft can also be terminated by diffusion out of the synapse, or via inactivation by catechol-O-methyltransferase
(COMT) (Kaakkola and Wurtman, 1992) and monoamine oxidase (MAO) (Brannan et al, 1995; Di Monte et al, 1996), respectively. COMT is primarily located in non-
dopaminergic cells and inactivates dopamine and other catecholamines by methylation.
MAO is an enzyme situated intraneuronally and elsewhere that inactivates
catecholamines and other amines by deamination (Goldstein and Lieberman, 1992).
Whereas reuptake presumably dominates removal of dopamine in the striatum, COMT is likely to have a more important role in the regulation of dopamine transmission in the cortex. In line with this, fMRI studies have shown that a DAT VNTR polymorphism has an effect on mesencephalic, but not on prefrontal, activity during episodic memory processing, whereas the COMT Val158Met genotype predicts prefrontal but not mesencephalic activity (Forbes et al, 2009; Schott et al, 2006).
Dopamine exerts its action by binding to one of five different dopamine receptors. These receptors belong to the family of seven transmembrane domain G protein-coupled receptors, and are divided into two subfamilies based on pharmacological and
biochemical properties. The D1 subfamily includes the D1 and D5 receptors, while the D2
subfamily includes D2, D3 and D4 receptors. Most dopamine agonists cannot strictly differentiate between members of the same subfamily (Girault and Greengard, 2004;
Vallone et al, 2000).
The D1 subfamily stimulates, whereas the D2 family inhibits, adenylyl cyclase and its second messenger, cyclic adenosine triphosphate (cAMP). The D2 receptor exists in two isoforms generated by alternative splicing of the same gene, one version being 29 amino acids shorter. The short and long isoforms are believed to be functionally different. The short version may function as an autoreceptor, regulating dopamine synthesis and release, whereas the long version, like the D1 receptor, is located postsynaptically (Lindgren et al, 2003; Usiello et al, 2000; Wang et al, 2000).
Dopamine c ell groups in the CNS
The different dopaminergic neuronal cell groups in the CNS are involved in controlling or modulating various parts of the brain. This thesis focuses on genes of importance for the development of mesodiencephalic dopaminergic (mdDA) neurons, which are made up of anatomically and functionally heterogeneous populations that are separate from other dopaminergic subgroups located elsewhere in the CNS. Below is a brief summary of the dopaminergic structures that can be identified in the murine brain.
Early studies of neuronal pathways in the CNS identified catecholamine cell groups, which were named A1 to A17. The A11-A15 groups of dopaminergic neurons are found in the diencephalon. The posterior hypothalamus cell group (A11) and the zona incerta cell group (A13) in the ventral thalamus are the largest dopaminergic cell groups in this area. The A11 group sends major projections to the spinal cord and lower brain stem, the function of which is poorly understood (Bjorklund and Skagerberg, 1979). The A13 group has more diffuse projections to the amygdala as well as different areas of the hypothalamus. The A12 (arcuate nucleus) and A14 (para- and periventricular
hypothalamic nucleus) groups project to the median eminence of the hypothalamus and the pituitary gland, providing the projections involved in neuroendocrine regulation mainly of prolactin release (Ben-Jonathan and Hnasko, 2001). The connectivity of the A15 group (lateral and ventral hypothalamus) is somewhat unclear. However, dopamine input to magnocellular neurons in the hypothalamic supraoptic nucleus is derived from the population of neurons located in the A14 and A15 cell groups (van Vulpen et al, 1999). The A16 and A17 groups consist of a small dopamine population in the
telencephalon. The A16 group of the olfactory bulb makes locally restricted connections as periglomerular interneurons and the A17 group comprises retinal amacrine
interneurons (Prakash and Wurst, 2006b).
Group A8-A10 are located in the mesencephalon. Evidence from molecular and electrophysiological studies suggests that mesencephalic dopaminergic neurons can be divided into three different subgroups; the retrorubral field (RRF), the ventral tegmental area (VTA) and substantia nigra pars compacta (SNc). These nuclei have historically been designated A8, A9 and A10 respectively. Three major pathways originate from the mdDA neurons: the mesolimbic pathway connects the ventral tegmental area (VTA) with the ventral striatum (nucleus accumbens, amygdala and olfactory tubercle), the mesocortical connects the VTA with the frontal cortex, and the nigrostriatal pathway
projects from the substantia nigra to the dorsal striatum (caudate nucleus and putamen).
However, the exact projection pattern of the dopamine neurons originating in the mesencephalon is a matter of debate. A recent study that used retrograde tracing
technique in mesocorticolimbic dopaminergic neurons from mice showed that dopamine projections in the medial prefrontal cortex, the basolateral amygdala and the core and medial shell of the nucleus accumbens originate in the medial posterior part of the VTA, whereas dopamine projections to the lateral shell of the nucleus accumbens originate in the more lateral portions of the VTA and the medial part of the SNc (Lammel et al, 2008). Furthermore, dopamine projections from the medial posterior VTA had low DAT mRNA and protein levels relative to their levels of TH and vesicular monoamine
transporter (Vmat2), whereas dopaminergic neurons originating in the lateral VTA to the nucleus accumbens lateral shell displayed high DAT expression (Lammel et al, 2008).
The mdDA neurons are involved in reward-related behaviour (Schultz, 2001), as well as stimulus-incentive learning (A10 neurons in particular) (Liu et al, 2008), controlling voluntary movements and regulating emotion-related behaviour and have been linked to various disorders, such as schizophrenia and Parkinson’s disease.
T RANSCRIP TI ONAL CONTRO L OF DOPAMINERGIC NEURON DEVEL O P M ENT
Early embryogenesi s
The development, properties and fate of mdDA neurons are ultimately controlled at the transcriptional level. During embryogenesis, three germ layers are formed: the endoderm, mesoderm and ectoderm. During gastrulation, neural progenitors are formed from the ectodermal layer after induction by signals from the dorsal mesoderm (Moreau and Leclerc, 2004). After this induction, the neuroectoderm grows thicker and closes-up to form the neural tube (Greene and Copp, 2009) at which time the forebrain, midbrain, hindbrain and spinal cord start to form along the anterior-posterior axis with the help of signal gradients and transcription factor domains (Lee and Pfaff, 2001; Lumsden and Krumlauf, 1996).
Development of mdDA neurons involves the activation of transcriptional cascades that have only been partially identified as of yet. In mice, the transcription factor Gbx2 is expressed in all three germ layers, but later restricted to the anterior hindbrain (Millet et al, 1999). Another transcription factor, Otx2 is first expressed in the epiblast, and later (at E7 in mouse) limited to the anterior neuroectoderm that will give rise to the forebrain and midbrain (Broccoli et al, 1999; Simeone et al, 2002). In Gbx2-/- knockout mice, the Otx2 expression pattern is moved posteriorly, leading to a faulty posterior expansion of anterior brain regions (Martinez-Barbera et al, 2001; Millet et al, 1999).
Mice in which Otx2 has been conditionally deleted at E9.5 display a severe decrease in mdDA neurons (Puelles et al, 2004). Otx2-/- knockout mice display even greater
abnormalities, failing to develop forebrain and midbrain (Acampora et al, 1995). The additional deletion of the homeodomain protein Nkx2.2 in conditional Otx2 knockout mice reverses the mdDA neuronal deficiency, thus suggesting that Nkx2.2 is a negative regulator of mdDA development, and that Otx2 normally represses Nkx2.2 (Prakash et al, 2006a).
The transcription factors En1/2, Foxa1/2 and Pax2/5 are activated shortly after
induction of Otx2 and Gbx2, around E8 in mice (Alavian et al, 2008; Ferri et al, 2007).
En1/2, Foxa1/2 and Pax2/5 participate in the regional specification of the midbrain and hindbrain and are essential for the development of mdDA neurons, as demonstrated by studies of knockout mutant mice (Liu and Joyner, 2001; McMahon and Bradley, 1990;
Schwarz et al, 1997; Sgado et al, 2008).
The secreted glycoproteins Wnt1 and Wnt5a regulate cell proliferation, fate decisions, and differentiation (Castelo-Branco et al, 2003). Studies suggest that Wnt1 and Wnt5a work in concert with other signals such as Shh and Fgf8 to promote mdDA neuronal development by establishing the dopaminergic progenitor domain in the mammalian ventral midbrain (Ye et al, 1998). Wnt1 has been suggested to be involved in the regulation of transcription factors Otx2 and Nkx2.2 (Prakash et al, 2006a; Prakash and Wurst, 2007), and for induction of the En1/2 genes (Danielian and McMahon, 1996).
Wnt1 increased the proliferation of Nurr1 (orphan nuclear hormone receptor Nr4a2) expressing precursors, whereas Wnt5a increased the proportion expressing Nurr1 and Pitx3, which are specifically expressed in mdDA neurons (Figure 1) (Castelo-Branco et al, 2003).
Formation of the isthmus and mdDA neuronal field
During development of the brain, neurons and other cells emerge at specific locations, partly controlled by local organizing centres that release inductive factors (Ye et al, 2001). The midbrain-hindbrain (mesencephalon-rhombencephalon) organizing centre, called isthmus, is established by mutual repression between transcription factors Otx2 and Gbx2 (Millet et al, 1999). After formation of the isthmus, signals from the floor plate and roof plate divide the mesencephalon and diencephalon into a dorsal and a ventral part, each governed by a unique transcriptional cascade (Smits et al, 2006). Each
organizing centre plays a unique role; the isthmus is essential for the location and size of mdDA neurons (Nakamura et al, 2008). Cells at this organizing centre produce Fgf8, which interacts with the glycoprotein Shh secreted by the floor plate, which runs throughout almost the entire length of the neural tube, to designate the specific place where mdDA neurons are formed in the ventral part of the mesencephalon and diencephalon (Hynes et al, 1995; Ye et al, 1998). Whereas the early onset of Fgf8 expression appears to be independent of En1 and En2 (Liu et al, 2001), these
transcription factors influence the size of Fgf8 expression in the isthmus, as demonstrated by animal studies (Shamim et al, 1999).
In addition to being expressed during embryo regionalization at E8 in mice, En 1 and En2 are also involved in later (i.e. E11-E12) specification of neuronal phenotype (Simon et al, 2004). Studies of En1/2-/- double knockout mice have hence demonstrated that these genes are required for survival of mdDA neurons (Wurst et al, 1994). Similarly, inactivation of these genes by RNA interference induces apoptosis (Simon et al, 2004).
Tgf is a factor that is required for survival of mdDA neurons; in vitro studies suggest that Tgf acts in cooperation with Shh and Fgf8 to induce TH expressing cells, but it does not appear to affect proliferation (Roussa et al, 2004). Tgf has been shown to reduce apoptosis in vitro and neutralization of this gene results in a reduced number of mdDA neurons (Farkas et al, 2003). Furthermore, Tgf induces glial cell line-derived
neurotrophic factor (GDNF), suggesting that the role of Tgf in mdDA neuron development and survival might rely on neurotrophic support (Peterziel et al, 2002).
Figure 1. The development of mdDA neurons requires a complex combination of diffusible signals and transcription factors in order to control both the acquirement and maintenance of a correct phenotype. Stem cells are first patterned to a ventral cell fate, after which they are further specified towards an mdDA neuronal fate by a transcriptional cascade comprising several stages.
Differentiati on of mdDA neurons requires Lmx1a
A transcriptional cascade involving several steps of specification controls the development of mature mdDA neurons (Alavian et al, 2008). After formation of the isthmus and the induction of Shh and Fgf8, around E8.5 in mouse, precursor cells
destined to become mdDA neurons begin to produce Lmx1a (Failli et al, 2002). Lmx1a is of great importance for mdDA neuronal development, since it alone is both sufficient (in conjunction with Shh) and required to trigger dopaminergic cell differentiation
(Andersson et al, 2006b; Thameem et al, 2002). Although Lmx1a is essential for the development of mdDA neurons, it is also expressed in other areas of the brain, including the hippocampus, cerebellum and the dorsal spinal cord (Failli et al, 2002).
Lmx1a induces the expression of Msx1, which in turn inhibits negative regulators of neurogenesis, including Nkx6.1 (Figure 1) (Andersson et al, 2006b). Consequently, Lmx1a and Msx1 function as determinants of mdDA neuron cell fate (Andersson et al, 2006b), initiating a transcriptional cascade that eventually induces the proneural protein Ngn2 and neuronal differentiation. In addition to Msx1, Otx2 also regulates Ngn2, as expression of the latter is absent in conditional Otx2 mutant knockout mice (Vernay et al, 2005). Ngn2-/- knockout mice have a reduced number of postmitotic mdDA neuronal markers, such as Nurr1 and TH (the rate-limiting enzyme in dopamine synthesis) (Andersson et al, 2006a), and Ngn2 is thus required for normal generation of mdDA neurons. Over-expression of Ngn2 in vitro results in increased neuronal differentiation and induces the expression of the mdDA neuronal marker Vmat2, but not Nurr1, indicating that Ngn2 is not sufficient for mdDA neuronal differentiation (Andersson et al, 2007; Roybon et al, 2008).
The spontaneously generated mutant dreher mouse (Sekiguchi et al, 1994; Sekiguchi et al, 1992) has been identified to carry a mutation in the Lmx1a gene (Millonig et al, 2000) and exhibits neurogenesis defects in mdDA neurons (Ono et al, 2007). Silencing the expression of Lmx1a using RNA interference in chick embryos also results in loss of mdDA neurons (Andersson et al, 2006b). Furthermore, forced expression of Lmx1a can under permissive conditions promote generation of mdDA neurons in mouse and human embryonic stem cells (Friling et al, 2009). After induction of Lmx1a expression, cells destined to become mdDA neurons gradually mature and become postmitotic.
Development of postmitotic mdDA neurons involves Lmx1b and Pitx3 Early phenotypic markers of postmitotic mdDA neurons, such as TH, are induced around E11.5 in mice (Abeliovich and Hammond, 2007). Transcription factors Nurr1, Lmx1b and Pitx3 are activated around this time. These genes are essential for
differentiation and survival of mdDA neurons (Nunes et al, 2003; Smidt et al, 2000;
Wallen and Perlmann, 2003).
The transcription factor Nurr1 is expressed immediately after mdDA neurons become postmitotic, briefly prior to TH induction at around E10.5 in mice (Wallen et al, 2003), and then stays expressed throughout adulthood (Backman et al, 1999). Nurr1-/- knockout mice fail to express TH in the SNc and VTA, but other markers such as Pitx3 and Lmx1b are unaltered (Castillo et al, 1998; Zetterstrom et al, 1997). Consequently, Nurr1 deficient mice do not fail to generate mdDA neurons, but because Nurr1 is responsible for the activation of TH, Vmat2, and DAT, these neurons lack proper dopaminergic phenotype (Smits et al, 2003). Later in life, Nurr1 has been suggested to play a role in survival and migration of mdDA neurons (Wallen et al, 1999), but this finding is disputed (Witta et al, 2000). Genetic studies in humans however have associated mutations in Nurr1 to PD (Jankovic et al, 2005), whereas studies in mice have linked Nurr1 deficiency to progressive loss of mdDA neurons in SNc and increased sensitivity to the
dopaminergic neurotoxin MPTP (Jiang et al, 2005; Le et al, 1999), suggesting that Nurr1 is indeed of importance for survival of some mdDA neurons. Interestingly, the
neurotrophin BDNF, which is also involved in the maturation of mdDA neurons, is a Nurr1 target gene; hence it is possible that some of the effects of Nurr1 on the development of mdDA neurons are mediated via regulation of BDNF expression (Volpicelli et al, 2007). Transcription factors Foxa1/2, first expressed at E8.0, are required for the expression of Nurr1 as well as Ngn2 (Ferri et al, 2007). Studies of
Foxa1/2 function in conditional mutant mice suggest that these transcription factors also regulate Lmx1a and Lmx1b expression and inhibit Nkx2.2 expression, in mdDA
progenitors (Lin et al, 2009).
Lmx1b is first expressed in midbrain at E7.5 and partly co-expressed with Lmx1a and Msx1 in mdDA neuronal progenitor cells (Andersson et al, 2006b), but also in other cells outside the midbrain (Smidt et al, 2000). Expression of Lmx1b is down-regulated at around E11 only to reappear in postmitotic neurons at E16. Expression of Lmx1b at this later stage is co-localized with Pitx3 and TH and continues throughout adulthood (Smidt et al, 2000). Lmx1b is structurally related to Lmx1a, but functions differently, not being able to induce mdDA cell fate. Experiments on chick embryos have shown that Lmx1b acts as an effector of the growth factor Fgf8 in the regulation of Wnt1 in developing mdDA neurons. Wnt1 expression is localized to the Lmx1b expression
domain. (Adams et al, 2000). Lmx1b-/- knockout mice display TH-positive neurons, but fail to express Pitx3, and mdDA neurons are subsequently lost at the time when the mice are born (Smidt et al, 2000).
Lmx1b may regulate an independent pathway necessary for expression of Pitx3 (Figure 1) (Smidt et al, 2000). Pitx3 is expressed exclusively in mdDA neurons shortly after
induction of Nurr1 at E11.5 (Smidt et al, 1997), and expression continues into
adulthood. Pitx3 appears to be required for the regulation of TH expression in mdDA neurons as well as for the generation and maintenance of these cells (Maxwell et al, 2005).
It has been suggested that Nurr1 and Pitx3 interact in regulating the dopaminergic pathway gene battery. A study of downstream target genes of Pitx3 revealed that expression of VMAT2 and DAT were greatly reduced in mdDA neurons of Pitx3- deficient mice (Hwang et al, 2009). Pitx3 has been suggested to directly activate
transcription of VMAT2 and DAT, and thereby contributing to function and/or survival of mdDA neurons. Since both VMAT2 and DAT are also known to be regulated by Nurr1 (Smits et al, 2003), it is possible that Pitx3 and Nurr1 cooperate in regulating mDA specification and maintenance, partly through an overlapping downstream pathway (Hwang et al, 2009).
In the absence of Pitx3, the Nurr1 transcriptional complex is kept in a repressed state by co-repressors, including SMRT and PSF, through recruitment of histone deacetylase (HDAC), which keep the target promoters de-acetylated and thus repressed.
Recruitment of Pitx3 induces the release of SMRT from the transcriptional complex, and hence initiates target gene transcription (Jacobs et al, 2009a; Jacobs et al, 2009b;
Martinat et al, 2006).
Moreover, Pitx3 is able to upregulate the expression of BDNF, which also has effects on mdDA neuron proliferation, survival and differentiation (Yang et al, 2008). BDNF has been shown to promote the survival of human fetal mdDA neurons in vitro (Studer et al, 1996). Conditional knockout elimination of BDNF in mice brain throughout postnatal development disrupts the organization of SNc dopaminergic neurons (Oo et al, 2009).
A naturally occurring blind mutant mouse strain, first discovered in 1968 and called aphakia, lacks part of the Pitx3 gene, making them Pitx3 deficient (Rieger et al, 2001;
Smidt et al, 1997). Aphakia mice initially express normal levels of TH but later develop a deficit of midbrain dopaminergic precursor cells that would later have matured into SNc neurons (Hwang et al, 2003; Semina et al, 2000; Smidt et al, 2004a; van den Munckhof et al, 2003). Interestingly, by birth, aphakia mice display a specific absence of dopaminergic neurons in the SNc as well as a loss of axonal projections to the dorsal striatum (Hwang et al, 2003; Semina et al, 2000; Smidt et al, 2004a; van den Munckhof et al, 2003). In contrast, dopaminergic neurons in the VTA are relatively unaffected by the lack of Pitx3 in aphakia mice.
Normally, Pitx3 is expressed in both VTA and SNc, but precedes TH in SNc, whereas it is concurrent with TH in the VTA (Maxwell et al, 2005; Messmer et al, 2007; Smidt et al, 1997). Studies suggest that the dependence of the SNc on Pitx3 is caused by SNc- specific activation of aldehyde dehydrogenase 2 (Ahd2) (Smidt and Burbach, 2009).
Pitx3 thus has been shown to act on the promoter of Ahd2 (Chung et al, 2005; Jacobs et al, 2007), which is expressed by dopaminergic neurons of the SNc and catalyses the formation of retinoic acid (RA). RA is a small signal molecule that has a crucial role in neuronal patterning, differentiation and survival of mdDA neurons (Jacobs et al, 2007;
McCaffery and Drager, 1994; Smidt et al, 2009). In vitro studies have shown that overexpression of Pitx3 increases the generation of dopaminergic neurons expressing Ahd2 (Chung et al, 2005; Martinat et al, 2006); moreover, restoration of RA signalling in the embryonic mdDA area counteracts the developmental defects caused by Pitx3 deficiency (Jacobs et al, 2007). Furthermore, in vitro studies have shown that Pitx3- deficient embryonic stem cells generate 50% fewer mature dopaminergic neurons, and that dopamine release was dysfunctional in these neurons (Papanikolaou et al, 2009).
The reduced number of generated cells was partially restored by the addition of RA, adding support for the notion that the effects of Pitx3 on mdDA neuron specification are mediated, at least in part, via the retinoic acid pathway (Papanikolaou et al, 2009).
Several of the transcription factors described above are known to play an important role in the maintenance and survival of mdDA neuron, as are numerous neurotrophic factors, including GDNF, conserved dopamine neurotrophic factor (CDNF), NRG1, Tgf/, Fgf2 and BDNF (Hyman et al, 1991; Lindholm et al, 2007; Peterziel et al, 2002; Roussa et al, 2004; Timmer et al, 2004; Yang et al, 2008; Yurek et al, 2004).
PARKINSON’S DISEASE
Bac kground, symptomat ology and tre atment
In his monograph published in 1817 entitled An Essay on the Shaking Palsy, James Parkinson described a disorder where patients display ”involuntary tremulous motion with lessened muscular power, in parts not in action even when supported, with a propensity to bend the trunk forward and to pass from a walking to a running pace”. In acknowledgement of his work, the famous neurologist Jean Martin Charcot later proposed that this primarily sporadic neurodegenerative disorder should be called Parkinson's disease (PD) (Kempster et al, 2007).
The incidence rate of PD rises considerably with age, with a lifetime risk of developing the disease of 1.5%, and with a median onset age of 60 years. Because of an aging population in the western world, PD is becoming more prominent and PD is now the second most frequent neurodegenerativedisorder after Alzheimer's disease (de Rijk et al, 1995; Lesage and Brice, 2009).
The term parkinsonism refers to the cardinal symptoms of PD, which include akinesia (inability to initiate movement), resting tremor (4–6 Hz), muscular rigidity and bradykinesia (slowness of voluntary movement with progressive reduction in speed and amplitude or repetitive actions). The onset of PD is gradual and the earliest symptoms might go unnoticed for some time. The symptoms mentioned above are typical but not unique to PD, and are often accompanied by other symptoms, including depression, dementia, and hyposmia (inability to detect odor). Other supportive criteria for PD include a unilateral onset with a progressive course and, initially, an excellent response to L-DOPA treatment (Lees et al, 2009).
